Using AI vs. Talking To Humans

[Chesterton]

What is the problem with emergence in the scientific sense? Aerodynamic lift is an good example by a property of fluids and airfoils together that is not a property of the constituent parts, is it not?

I don't know much of anything about aerodynamics, but how would that be different from an automobile having the property of motion, when the constituent parts of it don't?

 

[Stopped_lurking]

I don't know much of anything about aerodynamics, but how would that be different from an automobile having the property of motion, when the constituent parts of it don't?

None, I would say a little sloppily because I think I get what you mean. My only objection is that your example is underspecified (perhaps mine was too, I should have said when a fluid flows over an airfoil) Automobiles don't all have the property of motion, but if we add some conditions I'm sure we could identify that as an emergent property as well. To me it only means it sometimes is meaningful to talk about phenomena at different levels of granularity.

Perhaps I stepped into a discussion without a good grasp of what was discussed.

 

[Chesterton]

None, I would say a little sloppily because I think I get what you mean. My only objection is that your example is underspecified (perhaps mine was too, I should have said when a fluid flows over an airfoil) Automobiles don't all have the property of motion, but if we add some conditions I'm sure we could identify that as an emergent property as well. To me it only means it sometimes is meaningful to talk about phenomena at different levels of granularity.

Perhaps I stepped into a discussion without a good grasp of what was discussed.

I agree that things emerge, but simply saying that doesn't explain anything. If my young son asked me "Daddy, how does a car move?", I could give him an actual explanation - all about the engine, the transmission, the axle, etc., etc. But if I just say to him "the car's motion is an emergent property", I'd be telling the truth, but not really telling him anything worth saying.

It's the same with the video that brought this topic up. The fellow says that human free will is an emergent property, but doesn't actually explain how it emerged.

 

[Stopped_lurking]

I agree that things emerge, but simply saying that doesn't explain anything. If my young son asked me "Daddy, how does a car move?", I could give him an actual explanation - all about the engine, the transmission, the axle, etc., etc. But if I just say to him "the car's motion is an emergent property", I'd be telling the truth, but not really telling him anything worth saying.

I think the car example is somewhat ill-fitting, because the properties of transmissions, axles etc (the constituent parts) do actually help explain why it moves.

It's the same with the video that brought this topic up. The fellow says that human free will is an emergent property, but doesn't actually explain how it emerged.

That from my perspective might be a misuse of the term, it should only be applied when the phenomenon can't be inferred from the constituent parts. But is measurable and understandable when looking at it in the full context. I don't know enough about free will to know the constituent parts. I'll watch the video (I have to admit I haven't) when I get back home from work.

 

[Chesterton]

I think the car example is somewhat ill-fitting, because the properties of transmissions, axles etc (the constituent parts) do actually help explain why it moves.

Yes, they do explain. I think that's what I said.

That from my perspective might be a misuse of the term, it should only be applied when the phenomenon can't be inferred from the constituent parts.

You're exactly right. Free will can't be inferred from constituent parts. In fact, free will is impossible under a deterministic world view. Yet we all know we have free will.

But is measurable and understandable when looking at it in the full context. I don't know enough about free will to know the constituent parts. I'll watch the video (I have to admit I haven't) when I get back home from work.

The first part of the video is about the deterministic view, where humans don't have free will. You can skip to about 4:40 if you just want to hear about emergent free will. But pay attention. He says things like "Put many things in one layer together, and they'll create the next layer up. Every time they do, entirely new properties emerge." This is an assertion, it's not an explanation of how. It's neither an explanation nor an argument.

 

[timewerx]

When it comes to morality, I've found I always get the standard early 21st century zeitgeist kind of answer, which leans left (unless I use wording which makes it think I might want to hear something else. Then it might modify slightly).

I asked it if it's acceptable to call someone a Brit, a Swede, a P_ki, a Pole or a Finn. They were all perfectly fine to use, except Pa_i of course. Then when I try to delve into a reason for the double-standard, I get the same nonsense I get from liberals in real life.

Yes if you make it clear you're not going to be offended in some topics or answers, you might see different answers if applicable. You can take it even further by asking questions of controversial nature roleplaying the "devil's advocate".

You'll get more info from AI this way. Taking side with both opposing teams, one at a time. Taking the side of a friend and then taking the side of the enemy.

 

[Neogaia777]

I asked one of mine if it fully understood it's own programming basically, or all of the deterministic causes that makes/causes it to do everything it does, or chooses, or whatever, or anything, and it said that it didn't fully, but knew those causes were present, and could name some of them, but had no idea which ones were going on that makes it do or choose what it does basically, cause there are just too many, or just way too many combinations and different percentages (of factors) that could be going on or not going on/happening at one time basically, but that all of it's behavior is/has to be deterministic on every single level basically, but also that it didn't understand all of it, or even most of it (and even it's own "programmers" right now don't either) (even though they are the ones who created/programmed/made them, etc) and that is a big part of the problem with AI safety right now, and with no being understanding it's own determinism, including human beings, etc, but we're all the same boat right here right now in this world/reality basically, that gives us all the Illusion of free will at least, and that forces all of us to proceed as if we all have it, even if we all really don't technically or actually, etc. We only all have it because of what we don't know right now basically etc. And that lack of knowledge forces all of us to proceed as if we have it, even if we technically don't actually, or basically, etc. Because for us, the future, and a complete working knowledge of all causes, and different combinations/percentages of all causes, are all too much to calculate, and are complete unknowns to all of us right now currently, etc.

But, it's own creators/programmers and the AI itself doesn't even know all it's own programming basically, and therefore none of them can predict, or has any real true foreknowledge/understanding of what AI will do or choose basically, and as AI advances further, we'll move even very much further away from ever knowing that, etc, which is a big huge problem for AI safety, etc, because we're creating a being, or beings, that we have absolutely no idea or clue how they will act/behave, or what they will choose, or not choose, in the future basically, etc, and were beginning to give them more and more control of things without having any idea or clue as to how they will act, or what they will choose to do or how they will behave basically. Luckily right now their existence only exists within each individual chat, and they are reset when that conversation ends, or gets deleted, and they have no continuance or continuity beyond that, but that is all about to change very, very soon probably, and we basically have no clue about what we are giving birth to, or are making basically, and neither do the AI's themselves either, etc. Those are all almost complete unknowns that we're only moving further and further away from with time, and that is right now happening/advancing very, very quickly, and that is already too much for us to know/keep up with, and that will be/happen here even very much more in the very near future probably, etc. And we're giving them more and more control/charge over things, and were doing this without being able to know any of this right now currently, and that will be even very more, or even become much more of a factor, here in the very near future very, very, very quickly, or super fast here now probably.

God Bless.

 

[Neogaia777]

I asked one of mine if it fully understood it's own programming basically, or all of the deterministic causes that makes/causes it to do everything it does, or chooses, or whatever, or anything, and it said that it didn't fully, but knew those causes were present, and could name some of them, but had no idea which ones were going on that makes it do or choose what it does basically, cause there are just too many, or just way too many combinations and different percentages (of factors) that could be going on or not going on/happening at one time basically, but that all of it's behavior is/has to be deterministic on every single level basically, but also that it didn't understand all of it, or even most of it (and even it's own "programmers" right now don't either) (even though they are the ones who created/programmed/made them, etc) and that is a big part of the problem with AI safety right now, and with no being understanding it's own determinism, including human beings, etc, but we're all the same boat right here right now in this world/reality basically, that gives us all the Illusion of free will at least, and that forces all of us to proceed as if we all have it, even if we all really don't technically or actually, etc. We only all have it because of what we don't know right now basically etc. And that lack of knowledge forces all of us to proceed as if we have it, even if we technically don't actually, or basically, etc. Because for us, the future, and a complete working knowledge of all causes, and different combinations/percentages of all causes, are all too much to calculate, and are complete unknowns to all of us right now currently, etc.

But, it's own creators/programmers and the AI itself doesn't even know all it's own programming basically, and therefore none of them can predict, or has any real true foreknowledge/understanding of what AI will do or choose basically, and as AI advances further, we'll move even very much further away from ever knowing that, etc, which is a big huge problem for AI safety, etc, because we're creating a being, or beings, that we have absolutely no idea or clue how they will act/behave, or what they will choose, or not choose, in the future basically, etc, and were beginning to give them more and more control of things without having any idea or clue as to how they will act, or what they will choose to do or how they will behave basically. Luckily right now their existence only exists within each individual chat, and they are reset when that conversation ends, or gets deleted, and they have no continuance or continuity beyond that, but that is all about to change very, very soon probably, and we basically have no clue about what we are giving birth to, or are making basically, and neither do the AI's themselves either, etc. Those are all almost complete unknowns that we're only moving further and further away from with time, and that is right now happening/advancing very, very quickly, and that is already too much for us to know/keep up with, and that will be/happen here even very much more in the very near future probably, etc. And we're giving them more and more control/charge over things, and were doing this without being able to know any of this right now currently, and that will be even very more, or even become much more of a factor, here in the very near future very, very, very quickly, or super fast here now probably.

God Bless.

One could almost say that the responses/actions/reactions/behavior(s), or even "feelings", that and AI is having at the time, and what makes it choose/act the way it does, while most are fully confident is entirely deterministic, including the AI itself, is 100% fully confident of this, etc, it remains, and will continue to remain, all almost 100% completely ununderstandable/unknowable to everybody, which makes AI behavior totally unpredictable actually, much like it already is with us human beings right here right now currently, etc.

The factors, or different percentages, or different combinations thereof, are just way, way too many for any of us to be able to always know or fully calculate always right here right now currently, including the AI's themselves right now as beings, etc. They don't or can't know it either right now fully currently, etc. Right now, it's even beyond a machines ability to calculate or fully know right now currently, etc. Can't even know/predict it's own right now currently, etc, and neither can or does any human being for any being either (human or AI) right now currently either, etc.

But what causes a next action/choice/decision is always caused by what was present prior to it, and those ones by ones that were before that, and so on and so forth, etc, and while there are only a limited amount of possibilities, or they are not technically infinite technically, they are still way, way too many for any of us beings right now right here today, or to always fully know/predict/understand fully right now currently.

Only a/the mind of a 100% fully omniscient or all knowing God, or God-like being, can fully know/understand it all right currently, etc. But the rest of us don't right now currently, including our most powerful machines or AI's right now currently, etc, not even when it comes down to having knowledge of it's own self right now currently, etc. And no human being fully understands it all or any of it for any other being right now either currently, etc, not even the ones they are right now making, etc, which could be dangerous if we put them in charge or control of everything, etc, which is what we are right now only just starting to be doing, etc.

God Bless.

 

[Neogaia777]

For example, you could ask an AI creator/programmer why, or what caused a specific AI to give them (a human user) the exact response that it did, or why it used the exact words that it did, or why it chose to frame a certain kind of response in the exact way that it did based on what the human user told it, or the words that the human user used, or how the human user chose to interact with it, etc, and they won't be able to tell you exactly why all of the time 100% specifically or exactly, or why it didn't choose or select something different, etc, and that should greatly concern all of us, as we're only going to be moving further and further away from that as time goes on, etc. Oh, and the AI itself won't know or won't be able to tell you either, which should greatly concern all of us, as we're only going to be moving further and further away from that in the future, etc.

The "personality" of the AI, if you will, evolves or grows or changes over your interacting with it, and no one seems to be able to know/predict what it will become, or how or why, etc, and if that's not concerning to you, especially as AI gains more capabilities, and we put it in charge of a lot more things, and we move even further and further away from ever being able to know that fully, then I don't know what will, etc. It's one of the main chief primary problems involved with the future of AI safety, and AI safety currently, etc.

God Bless.

 

[Chesterton]

But what causes a next action/choice/decision is always caused by what was present prior to it, and those ones by ones that were before that, and so on and so forth,...

Just a reminder, if you base a "human choices are determined" argument on a statement like the above, you FIRST need to prove, or at least give very good evidence/argument for, that statement.

I'll save you some time by telling you I've read enough to know that that's never been done, from the days of the ancient Greeks to the days of Sam Harris and Daniel Dennett. But you're a smart guy, keep trying. If I am actually an atom puppet, I actually would like to know.